You are on page 1of 43

Application Note

Poly-Phase Energy Metering IC

M90E36A

APPLICATION OUTLINE

This document describes system application issues when using the M90E36A (poly-
phase energy metering ICs) to design poly-phase energy meters.
The M90E36A is applicable in class 0.5S or class 1 poly-phase meter design and also
supports three-phase four-wire (3P4W, Y0) or three-phase three-wire (3P3W, Y or )
connection modes. The M90E36A can also be used in harmonic meter design.
The M90E36A uses 3.3V single power supply. In a typical 3P4W design, there are three
transformers&regulators to provide power supply. The AC power supply outputs 3.3V to
chip digital power supply DVDD after rectifier and voltage regulation. The analog power
supply AVDD should be connected directly to digital power supply DVDD.
The M90E36A has on-chip power-on-reset circuit. The RESET pin should be connected to
DVDD through a 10k resistor and a 0.1F filter capacitor to ground. The M90E36A has
highly stable on-chip reference power supply. The Vref pin should be decoupled with a
10F capacitor and a 0.1F ceramic capacitor.
The M90E36A employs 16.384MHz as the system frequency. The M90E36A has built-in
crystal oscillator circuit and 10pF matching capacitance. Users only need to connect a
16.384MHz crystal between OSCI and OSCO pins in application.
The M90E36A provides a 4-wire SPI interface (CS, SCLK, SDI and SDO) for external MCU
connection. MCU can perform chip configuration and register reading/writing through SPI.
The M90E36A also supports Master mode SPI, which is named Direct Memory Access
(DMA) mode. In DMA mode, The M90E36A streams out ADC sampling raw data to
external MCU at an up to 1800kbps rate.
The M90E36A provides four energy pulse output pins: active energy pulse CF1, reactive
energy pulse CF2 (can also be configured as apparent energy pulse), fundamental energy
pulse CF3 and harmonic energy pulse CF4. They can be used for energy metering cali-
bration and can also be connected to MCU for energy accumulation.
The M90E36A provides three zero-crossing pins ZX0, ZX1 and ZX2 which can select
different phases voltage or current as inputs.
The M90E36A provides three output pins IRQ0, IRQ1 and WarnOut to generate interrupt
and warn out signals at different levels.
The default application in this document is 3P4W, otherwise it will be specially indicated.

Atmel-46104A-SE-M90E36A-ApplicationNote_050514
Ta bl e o f C o n t en ts

1 HARDWARE REFERENCE DESIGN ........................................................................................... 4


1.1 3P4W Application ................................................................................................................... 4
1.1.1 Schematics (Current Transformer (CT)) ................................................................................... 4
1.1.2 BOM (CT) ................................................................................................................................. 5
1.1.3 Schematics (Rogowski) ............................................................................................................ 6
1.1.4 BOM (Rogowski) ....................................................................................................................... 7
1.1.5 Circuit Description ..................................................................................................................... 7
1.2 3P3W Application ................................................................................................................... 9
1.2.1 Schematics ............................................................................................................................... 9
1.2.2 BOM ........................................................................................................................................ 10
1.2.3 Circuit Description ................................................................................................................... 10
2 INTERFACE ................................................................................................................................ 11
2.1 SPI ........................................................................................................................................ 11
2.2 DMA...................................................................................................................................... 12
3 POWER MODES ........................................................................................................................ 13
3.1 Normal Mode ........................................................................................................................ 13
3.2 Partial Measurement Mode................................................................................................... 13
3.3 Detection Mode..................................................................................................................... 13
3.4 Idle Mode .............................................................................................................................. 14
3.5 Transition and Application of Power Modes ......................................................................... 14
4 CALIBRATION ........................................................................................................................... 17
4.1 Calibration Method................................................................................................................ 17
4.2 Calibration in Normal Mode .................................................................................................. 17
4.2.1 Measurement/Metering Startup Command (Configstart/Calstart/HarmStart/AdjStart) ........... 18
4.2.2 PL Constant Configuration (PL_Constant) ............................................................................. 20
4.2.3 Metering Method Configuration (MMode0) ............................................................................. 21
4.2.4 PGA Gain Configuration (MMode1) ........................................................................................ 22
4.2.5 Offset Calibration of Voltage/ Current/ Power ......................................................................... 23
4.2.6 Voltage/ Current Measurement Calibration ............................................................................ 24
4.2.7 Energy Metering Calibration ................................................................................................... 25
4.2.8 Fundamental Energy Metering Calibration ............................................................................. 26
4.3 Calibration in Partial Measurement Mode............................................................................. 28
4.3.1 Partial Measurement Configuration (PMConfig) ..................................................................... 28
4.3.2 Sampling Cycle Configuration (PMAvgSamples) ................................................................... 28
4.3.3 PGAgain Configuration (PMPGA) ........................................................................................... 28
4.3.4 Current Offset Calibration ....................................................................................................... 29
4.3.5 Current Measurement Calibration ........................................................................................... 29
4.3.6 Special Application of Partial Measurement Function ............................................................ 30
4.4 Calibration in Detection Mode............................................................................................... 31
4.4.1 Current Detection Module Configuration ................................................................................ 31
4.4.2 Current Detection Threshold Calibration ................................................................................ 32

M90E36A [APPLICATION NOTE] 2


Atmel-46104A-SE-M90E36A-ApplicationNote_043014
5 FUNCTION REGISTERS CONFIGURATION ............................................................................ 33
5.1 Startup Current Configuration............................................................................................... 33
5.2 SAG Function ....................................................................................................................... 34
5.3 Reserved Register/ Address and Reserved bits................................................................... 35
5.3.1 Reserved Register/ Address ................................................................................................... 35
5.3.2 Reserved Register Bits ........................................................................................................... 35
5.3.3 Reserved Bits in the FUNC_EN1 Register ............................................................................. 35
6 TEMPERATURE COMPENSATION .......................................................................................... 36
6.1 On-chip Temperature Sensor Configuration......................................................................... 36
6.2 Temperature Compensation Based on ADC Sampling Channel.......................................... 36
6.3 Temperature Compensation Based on reference Voltage ................................................... 38
7 HARMONIC ANALYSIS ............................................................................................................. 39
7.1 DFT Engine........................................................................................................................... 39
7.2 Obtain Harmonic Analysis of Above 32nd ............................................................................ 42

M90E36A [APPLICATION NOTE] 3


Atmel-46104A-SE-M90E36A-ApplicationNote_043014
4
1

1.1
1 2 3 4 5 6 7 8

U1 PS2501 JP1
UA CF1 1
R3 R4 R5 R6 R7 R8 R9 1
VAP 2
2
240K 240K 240K 240K 240K 240K 240K 3
R10 3
4
1K C3 4
5
18nF R1 5
A CON-5 A
510
GND GND
C1 DVDD33
GND
U2 PS2501
R2
UB 0.1uF CF2
R16 R17 R18 R19 R20 R21 R22 10K
VBP C2

M90E36A [Application Note]


240K 240K 240K 240K 240K 240K 240K
3P4W APPLICATION

R23
1K C6 10uF R11 R12 R13 R14
18nF 10K 10K 10K 10K R15
C4 510

Atmel-46104A-SE-M90E36A-ApplicationNote_050514
GND GND
DVDD33 0.1uF
SDI
U3 PS2501
SDO
UC AVDD33 C5 CF3
R25 R26 R27 R28 R29 R30 R31 SCLK
VCP 0.1uF
GND CS
240K 240K 240K 240K 240K 240K 240K
R32 C7

48
47
46
45
44
43
42
41
40
39
38
37
1K C8 0.1uF
18nF R24

CS

NC
NC
SDI
RST
VPP
SDO
510

SCLK

DGND
DGND
GND GND

DVDD33
DVDD18
1 36
1.1.1 Schematics (Current Transformer (CT))

B AVDD33 DMA DMA B


2 35
AGND NC
IAP 3 34 U4 PS2501
I1P PM1 PM1
UN IAN 4 33 CF4
I1N PM0 PM0
Voltage Sampling IBP 5 32
I2P TEST GND
IBN 6 31
I2N IRQ1 IRQ1
ICP 7 30
I3P IRQ0 IRQ0
GND ICN 8 29
I3N WarnOut WarnOut R33
HARDWARE REFERENCE DESIGN

INP 9 28 CF4
I4P CF4 CF4
INN 10 27 CF3 510
I4N CF3 CF3
11 26 CF2
Vref CF2 CF2
12 25 CF1 GND
C9 C10 AGND CF1 CF1
10uF 0.1uF

V1P
V1N
V2P
V2N
V3P
V3N
NC
OSCI
OSCO
ZX0
ZX1
ZX2
GND
U5 Energy Pulse Output (Isolated with Optocoupler)

13
14
15
16
17
18
19
20
21
22
23
24
ATM90E36A
ZX2
ZX1
ZX0
IA+ IC+

VAP
VBP
VCP
R35 IAP R36 ICP
X1
1K 1K R34
R38 R39 16.384MHz Connect to MCU CF1
2.4 C11 2.4 C12
510
18nF 18nF D1 CF4
R40
C 1K R41 C
C13 C14 R43 C15
1K 18nF 1K 18nF R37
18nF CF2
R44 GND C16 R45 GND C17
510
IA- 2.4 18nF IC- 2.4 18nF D2 CF3
R46 IAN R47 ICN
1K 1K GND
CF3 R42
510
D3 CF2

IB+ IN+ CF4 R48


R49 IBP R50 INP Poly Phase Metering AFE Chip (ATM90E36A) 510
1K 1K D4 CF1
R51 R52 GND
2.4 C18 2.4 C19
18nF 18nF

Energy Pulse Output Indicate


R53 GND C20 R54 GND C21
IB- 2.4 18nF IN- 2.4 18nF
R55 IBN R56 INN
1K 1K Atmel China SMART ENERGY

DrawnBy: Felix Yao


D Option for N-line current sampling CheckedBy: * D

Current Sampling (with CT) ApprovedBy: * Size: A3

Document Number: * Page: 1 of 1

Date: 3/19/2014 10:59:16 Revision: 1.0

Project: * File: ATM90E36A_3P4W_CT.SchDoc

Title: Poly Phase Metering AFE ATM90E36A (3P4W with CT)

1 2 3 4 5 6 7 8
1.1.2 BOM (CT)
Table-1 3P4W BOM (CT)
Component Type Designator Quantity Parameter Tolerance
SMT Capacitor C3 C6 C8 C11 C12 C13 14 18nF 10% X7R
C14 C15 C16 C17 C18 (anti-aliasing filter capacitor)
C19 C20 C21
C1 C4 C5 C7 C10 5 0.1F 10% X7R
C2 C9 2 10F 10% X7R
SMT Resistor R38 R39 R44 R45 R51 8 2.4 1% 1/8W 25ppm
R52 R53 R54
R1 R15 R24 R33 R34 8 510 5% 1/8W 100ppm
R37 R42 R48
R10 R23 R32 R35 R36 14 1k 1% 1/8W 25ppm
R40 R41 R43 R46 R47 (anti-aliasing filter resistor)
R49 R50 R55 R56
R2 R11 R12 R13 R14 5 10k 5% 1/8W 100ppm
R3~R9, R16~R22, 21 240k 1% 1/8W 25ppm
R25~R31
LED D1 D2 D3 D4 4 - -
SMT Optocoupler U1 U2 U3 U4 4 PS2501 -
Crystal X1 1 16.384MHz 20ppm
IC U5 1 M90E36A -
Connector JP1 1 CON-5 -

M90E36A [APPLICATION NOTE] 5


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
6
1 2 3 4 5 6 7 8

U1 PS2501 JP1
UA CF1 1
R3 R4 R5 R6 R7 R8 R9 1
VAP 2
2
240K 240K 240K 240K 240K 240K 240K 3
R10 3
4
1K C3 4
5
18nF R1 5
A CON-5 A
510
GND GND
C1 DVDD33
GND
U2 PS2501
R2
UB 0.1uF CF2
R16 R17 R18 R19 R20 R21 R22 10K
VBP C2
1.1.3 Schematics (Rogowski)

M90E36A [Application Note]


240K 240K 240K 240K 240K 240K 240K
R23
1K C6 10uF R11 R12 R13 R14
18nF 10K 10K 10K 10K R15
C4 510

Atmel-46104A-SE-M90E36A-ApplicationNote_050514
GND GND
DVDD33 0.1uF
SDI
U3 PS2501
SDO
UC AVDD33 C5 CF3
R25 R26 R27 R28 R29 R30 R31 SCLK
VCP 0.1uF
GND CS
240K 240K 240K 240K 240K 240K 240K
R32 C7

48
47
46
45
44
43
42
41
40
39
38
37
1K C8 0.1uF
18nF R24

CS

NC
NC
SDI
RST
VPP
SDO
510

SCLK

DGND
DGND
GND GND

DVDD33
DVDD18
1 36
B AVDD33 DMA DMA B
2 35
AGND NC
IAP 3 34 U4 PS2501
I1P PM1 PM1
UN IAN 4 33 CF4
I1N PM0 PM0
Voltage Sampling IBP 5 32
I2P TEST GND
IBN 6 31
I2N IRQ1 IRQ1
ICP 7 30
I3P IRQ0 IRQ0
GND ICN 8 29
I3N WarnOut WarnOut R33
INP 9 28 CF4
I4P CF4 CF4
INN 10 27 CF3 510
I4N CF3 CF3
11 26 CF2
Vref CF2 CF2
12 25 CF1 GND
C9 C10 AGND CF1 CF1
10uF 0.1uF

V1P
V1N
V2P
V2N
V3P
V3N
NC
OSCI
OSCO
ZX0
ZX1
ZX2
GND
U5 Energy Pulse Output (Isolated with Optocoupler)

13
14
15
16
17
18
19
20
21
22
23
24
ATM90E36A
ZX2
ZX1
ZX0
IA+ IC+

VAP
VBP
VCP
R38 R35 IAP R39 R36 ICP
X1
1K 1K 1K 1K R34
16.384MHz Connect to MCU CF1
C22 C11 C23 C12
510
18nF 18nF 18nF 18nF D1 CF4
R40
C 1K R41 C
C13 C14 R43 C15
1K R37
18nF 18nF 1K 18nF CF2
C24 GND C16 C25 GND C17
510
IA- 18nF 18nF IC- 18nF 18nF D2 CF3
R44 R46 IAN R45 R47 ICN
1K 1K 1K 1K GND
CF3 R42
510
D3 CF2

IB+ IN+ CF4 R48


R51 R49 IBP R52 R50 INP Poly Phase Metering AFE Chip (ATM90E36A) 510
1K 1K 1K 1K D4 CF1
GND
C26 C18 C27 C19
18nF 18nF 18nF 18nF

Energy Pulse Output Indicate


C28 GND C20 C29 GND C21
IB- 18nF 18nF IN- 18nF 18nF
R53 R55 IBN R54 R56 INN
1k 1K 1K 1K Atmel China SMART ENERGY

DrawnBy: Felix Yao


D Option for N-line current sampling CheckedBy: * D

Current Sampling (with Rogowski Coil) ApprovedBy: * Size: A3

Document Number: * Page: 1 of 1

Date: 3/19/2014 10:59:07 Revision: 1.0

Project: * File: ATM90E36A_3P4W_Coil.SchDoc

Title: Poly Phase Metering AFE ATM90E36A (3P4W with Rogiwski Coil)

1 2 3 4 5 6 7 8
1.1.4 BOM (Rogowski)
Table-2 3P4W BOM (Rogowski)
Component Type Designator Quantity Parameter Tolerance
SMT Capacitor C3 C6 C8 C11 C12 C13 22 18nF 10% X7R
C14 C15 C16 C17 C18 (anti-aliasing filter capacitor)
C19 C20 C21 C22 C23
C24 C25 C26 C27 C28
C29
C1 C4 C5 C7 C10 5 0.1F 10% X7R
C2 C9 2 10F 10% X7R
SMT Resistor R1 R15 R24 R33 R34 8 510 5% 1/8W 100ppm
R37 R42 R48
R10 R23 R32 R35 R36 14 1k 1% 1/8W 25ppm
R38 R39 R40 R41 R43 (anti-aliasing filter resistor)
R44 R45 R46 R47 R49
R50 R51 R52 R53 R54
R55 R56
R2 R11 R12 R13 R14 5 10k 5% 1/8W 100ppm
R3~R9, R16~R22, 21 240k 1% 1/8W 25ppm
R25~R31
LED D1 D2 D3 D4 4 - -
SMT Optocoupler U1 U2 U3 U4 4 PS2501 -
Crystal X1 1 16.384MHz 20ppm
IC U5 1 M90E36A -
Connector JP1 1 CON-5 -

1.1.5 Circuit Description


The recommended circuit for the M90E36A three-phase four-wire (3P4W) application is as shown in 1.1.1 Schematics
(Current Transformer (CT)). The M90E36A can use CT and Rogowski coil in current sampling. The recommended circuit
for 3P4W application with Rogowski coil is as shown in 1.1.3 Schematics (Rogowski).
It is recommended to use two-order filtering when sampling with Rogowski coil. The other parts are the same as the CT
application circuit. The recommended type of Rogowski coil is: PA3202NL (Pulse Electronics).
Poly-phase voltage is sampled over resistor divider network with recommended ratio of 240K x 7:1K. The anti-aliasing
filter capacitor is recommended to be 18nF. Poly-phase current and N line current are sampled over CT. The CT ratio and
load resistance should be selected based on the actual metering range. The anti-aliasing filter resistance/capacitor is
suggested to be 1K/18nF for the current sampling circuit.
The CF1~CF4 pins are provided with driving capacity of 8mA which can drive LED and optocoupler parallelly. The other
digital pins are provided with driving capacity of 3mA which can drive optocoupler directly.

M90E36A [APPLICATION NOTE] 7


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
Application note: how to select CT and CT load resistance
Condition:
M90E36A ADC input voltage range is 120Vrms ~ 720mVrms
M90E36A ADC input gain PGA_GAIN = 1, 2, 4
Assume:
Metering range of the energy meter is Imin ~ Imax
CT current output ratio is N:1
CT load resistance is RCT
So the parameters meet the formula as below:
PGA_GAIN RCT I min
120 Vrms <
N

PGA_GAIN RCT I max


< 720mVrms
N

8 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
1.2

1 2 3 4 5 6 7 8
1.2.1 Schematics

U1 PS2501 JP1
UA CF1 1
R3 R4 R5 R6 R7 R8 R9 1
VAP 2
2
240K 240K 240K 240K 240K 240K 240K 3
R10 3
4
1K C3 4
5
18nF R1 5
A CON-5 A
510
GND GND
C1 DVDD33
GND
U2 PS2501
R2
UB 0.1uF CF2
10K
C2
3P3W APPLICATION

GND 10uF R11 R12 R13 R14


10K 10K 10K 10K R15
C4 510
GND
DVDD33 0.1uF
SDI
U3 PS2501
SDO
UC AVDD33 C5 CF3
R25 R26 R27 R28 R29 R30 R31 SCLK
VCP 0.1uF
GND CS
240K 240K 240K 240K 240K 240K 240K
R32 C7

48
47
46
45
44
43
42
41
40
39
38
37
1K C8 0.1uF
18nF R24

CS

NC
NC
SDI
RST
VPP
SDO
510

SCLK

DGND
DGND
GND GND

DVDD33
DVDD18
1 36
B AVDD33 DMA DMA B
2 35
AGND NC
IAP 3 34 U4 PS2501
I1P PM1 PM1
IAN 4 33 CF4
I1N PM0 PM0
Voltage Sampling IBP 5 32
I2P TEST GND
IBN 6 31
I2N IRQ1 IRQ1
ICP 7 30
I3P IRQ0 IRQ0
GND ICN 8 29
I3N WarnOut WarnOut R33
INP 9 28 CF4
I4P CF4 CF4
INN 10 27 CF3 510
I4N CF3 CF3
11 26 CF2
Vref CF2 CF2
12 25 CF1 GND
C9 C10 AGND CF1 CF1
10uF 0.1uF

V1P
V1N
V2P
V2N
V3P
V3N
NC
OSCI
OSCO
ZX0
ZX1
ZX2
GND
U5 Energy Pulse Output (Isolated with Optocoupler)

13
14
15
16
17
18
19
20
21
22
23
24
ATM90E36A
ZX2
ZX1
ZX0
IA+ IC+

VAP
VBP
VCP
R35 IAP R36 ICP
X1
1K 1K R34
R38 R39 16.384MHz Connect to MCU CF1
2.4 C11 2.4 C12
510
18nF 18nF D1 CF4
R40
C 1K C
C13 R43 C15
18nF 1K 18nF CF2 R37
R44 GND C16 R45 GND C17
510
IA- 2.4 18nF IC- 2.4 18nF D2 CF3
R46 IAN R47 ICN
1K 1K GND
CF3 R42
510
D3 CF2

CF4 R48
Poly Phase Metering AFE Chip (ATM90E36A) 510
D4 CF1
GND

Energy Pulse Output Indicate

Atmel China SMART ENERGY

DrawnBy: Felix Yao


D CheckedBy: * D

Current Sampling (with CT) ApprovedBy: * Size: A3

Document Number: * Page: 1 of 1

Date: 3/19/2014 10:58:57 Revision: 1.0

Project: * File: ATM90E36A_3P3W_CT.SchDoc

Title: Poly Phase Metering AFE ATM90E36A (3P3W with CT)

1 2 3 4 5 6 7 8

M90E36A [APPLICATION NOTE]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
9
1.2.2 BOM
Table-3 3P3W BOM
Component Type Designator Quantity Parameter Tolerance
SMT Capacitor C3 C8 C11 C12 C13 8 18nF 10% X7R
C15 C16 C17 (anti-aliasing filter capacitor)
C1 C4 C5 C7 C10 5 0.1F 10% X7R
C2 C9 2 10F 10% X7R
SMT Resistor R38 R39 R44 R45 4 2.4 1% 1/8W 25ppm
R1 R15 R24 R33 R34 8 510 5% 1/8W 100ppm
R37 R42 R48
R10 R32 R35 R36 R40 8 1k 1% 1/8W 25ppm
R43 R46 R47 (anti-aliasing filter resistor)
R2 R11 R12 R13 R14 5 10k 5% 1/8W 100ppm
R3~R9, R25~R31 14 240k 1% 1/8W 25ppm
LED D1 D2 D3 D4 4 - -
SMT Optocoupler U1 U2 U3 U4 4 NEC2501 -
Crystal X1 1 16.384MHz 20ppm
IC U5 1 M90E36A -
Connector JP1 1 CON-5 -

1.2.3 Circuit Description


This circuit is the recommended circuit for the M90E36A three-phase three-wire (3P3W) application.
Phase B is the reference ground in 3P3W application. In 3P3W system, Uab stands for Ua, Ucb stands for Uc and there is
no Ub.
Phase B voltage, phase B current and N line sampling current are not needed in 3P3W application. Pin 5, 6, 9, 10, 15 and
16 should be connected to GND.
If DMA function is not used, pin 36 should also be connected to GND. All NC pins should be left open.
The other parts of 3P3W application circuit are similar to 3P4W and can be treated in the same way.

10 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
2 INTERFACE
The M90E36A provides a four-wire SPI interface (CS, SCLK, SDI and SDO). The interface can be configured to two modes
by the DMA_CTRL pin: Slave mode and Master mode.

2.1 SPI
The SPI interface in Slave mode is mainly used for register read/write operation. A complete SPI read/write operation is of
32 bits, which contains 16-bit address and 16-bit data. In the 16-bit address, bit0 ~ bit9 correspond to valid register address
A0 ~ A9, and bit10 ~ bit14 are reserved (these bits are dont-care). Bit15 indicates the SPI operation is read or write.

SPI Operation Description Highest Bit (Bit15)


Read Read register data 1
Write Write data to register 0

The transmission of address and data bits is from high to low, which means MSB first and LSB last. Note that the M90E36A
read/write only supports single address operation, rather than continuous read or write.
The M90E36A has a special register LastSPIData [0FH] for recording the last SPI read/write data. This register can be
used for data check for SPI read/write operation. When the system is in strong interference situation, the disturbance signal
may cause SPI communication disorder and result in SPI read/write error. In this case, LastSPIData can be used to check
the correctness of SPI read/write and strengthen system robustness. For read-clear registers, if the read data is different
from the LastSPIData data, the actual data can be obtained by reading the LastSPIData register repeatedly.
LastSPIData application is as shown in Figure-1 and Figure-2:

SPI Read Data


Buffer

Read LastSPIData

LastSPIData == Buffer ?
Y

Buffer=LastSPIData

Read LastSPIData

LastSPIData == Buffer ?
N
Y

End

Figure-1 LastSPIData Application (Read)

M90E36A [APPLICATION NOTE] 11


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
SPI Write Data
Buffer

Read LastSPIData

N
LastSPIData == Buffer ?

End

Figure-2 LastSPIData Application (Write)

2.2 DMA
For details please refer to the SPI/DMA Interface chapter in the M90E36A datasheet.

12 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
3 POWER MODES
Four power modes are supported which correspond to four kinds of power consumption. The power mode is configured by
PM1/PM0 pins.

PM1 PM0 Power Modes Power Consumption


1 1 Normal mode High
1 0 Partial Measurement mode
0 1 Detection mode Low
0 0 Idle mode

3.1 NORMAL MODE


In Normal mode, all function blocks are active except for the current detector block. All registers can be accessed, including
the registers related to Partial Measurement mode and Detection mode.

3.2 PARTIAL MEASUREMENT MODE


In Partial Measurement mode, only three-phase current sampling and the related blocks are active. SPI communication is
normal in this mode, but only partial measurement related registers and some special registers can be accessed by
external MCU.The accessible registers in Partial Measurement mode are listed as below:

Address Name Address Name Address Name


00H SoftReset 14H PMOffsetA 1AH PMIrmsC
01H SysStatus0 15H PMOffsetB 1BH PMConfig
03H FuncEn0 16H PMOffsetC 1CH PMAvgSamples
07H ZXConfig 17H PMPGA 1DH PMIrmsLSB
0EH DMACtrl 18H PMIrmsA
0FH LastSPIData 19H PMIrmsB

There is a special enable control bit ReMeasure (bit14 of PMConfig) for Partial Measurement mode. When the control bit is
enabled, sampling and measurement are proceeded at the sampling period determined by the PMAvgSamples[1CH]
register.
Measure function is automatically shut off upon measurement completion. It needs to be enabled again if to measure
again. Upon measurement completion, the IRQ0 pin outputs high level. MCU can judge whether measurement is
completed through IRQ0. IRQ0 is cleared when the control bit (ReMeasure) is enabled again or partial measurement mode
is exited.
There is also a special Busy indication bit PMBusy (bit0 of PMConfig) for Partial Measurement mode. MCU can also
judge whether measurement is completed through the PMBusy bit.
Accuracy of current measurement in Partial Measurement mode is the same as Normal mode, because reference power
supply module is active.

3.3 DETECTION MODE


In Detection mode, only the current detector is active and all the registers can not be accessed by external MCU. In this
mode, each I/O is in specific state (for details refer to datasheet) and SPI is disabled. So the control and threshold registers
for Detection mode need to be programmed in Normal mode before entering Detection mode. Once these related registers
are written, there is no need to re-configure them when switching between different power modes. Detection mode related
registers are listed as below:

M90E36A [APPLICATION NOTE] 13


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
Address Name
10H DetectCtrl
11H DetectThA
12H DetectThB
13H DetectThC

Current detection is achieved with low power comparators. Two comparators are supplied for each phase on detecting
positive and negative current. When any single-phase current or multiple-phase current exceeds the configured threshold,
the IRQ0 pin is asserted high. When all three phase currents exceed the configured threshold, the IRQ1 pin is asserted
high. The IRQ0/IRQ1 state is cleared when entering or exiting Detection mode.
The all three phase currents are considered as the currents of three current channels I1~I3. As there is no phase B current
in 3P3W application, IRQ1 will not be asserted high even if both phase A and phase C current exceed the configured
threshold.

3.4 IDLE MODE


In Idle mode, all the modules are disabled and all the registers can not be accessed. In this mode, each I/O is in a specific
state (for details refer to datasheet) and SPI is disabled. All register values are lost except for current detection related
registers.

3.5 TRANSITION AND APPLICATION OF POWER MODES


The four power modes are controlled by the PM0 and PM1 pins. In application, any power mode transition goes through
Idle mode to avoid register value confusion or system status uncertainty in mode transition. All function modules are
disabled in Idle mode while the related modules will be enabled after switching from Idle mode to other mode, which is
equivalent to reset to the function modules, thus ensuring normal operation of the function modules.
It needs to reload registers to ensure normal operation when switching from Idle mode to Normal mode or Partial
Measurement mode, while no need to reload registers when switching from Idle mode to Detection mode. Power mode
transition is shown as Figure-3:

Normal Mode Need to reload all


PM1:PM0 = 11 register values

All the register


values will be lost
Idle Mode
except for the
PM1:PM0 = 00
Detection mode
related registers

Partial Measurement
Detection Mode
Mode
PM1:PM0 = 01
PM1:PM0 = 10
Need to reload Partial
Detection Mode related Measurement related
register value will be kept registers
Figure-3 Power Mode Transition

14 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
Note: For description convenience, the intermediary Idle mode will be omitted when refering to power mode transition.
The most typical application of power mode transition is no-voltage detection for power meter.
The so-called no-voltage state is when all phase voltages are less than the voltage threshold but the load current is greater
than the configured current value (such as 5% of rated current). In no-voltage state, the power meter usually uses backup
battery for power supply. The system needs to enter low power mode and perform measurement and recording for no-
voltage state periodically.
The recommended flow for power meter with the M90E36A is as below:
1 Set the current detection threshold to be the minimum load current (such as 5% of rated current) required in no-voltage
state.
2 When no-voltage happens, the system enters Idle mode;
3 The system enters Detection mode every once in a while (such as 5s);
4 Once the load current is greater than the configured value, the system enters Partial Measurement mode to measure
and record the load current;
5 The system returns to Idle mode after measurement and recording are completed;
6 The system enters Partial Measurement mode every once in a while (such as 60s) to measure and record the load
current.

System voltage sag


Enter Partial
Measurement mode

Enter Idle mode Measure


N
current>minimum load
current?
Y
Delay 5s
Record current
value
(no-voltage event)
Enter Detection
mode

Enter Idle mode

N
IRQ0/IRQ1 output high
level?
Delay 60s
Y

Figure-4 Application of Detection Mode and Partial Measurement Mode

M90E36A [APPLICATION NOTE] 15


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
Application note: Design principle for current detection threshold
It is recommended to do system design based on current detection threshold of 3mVrms.
Example:
Assume:
The requirement is that the minimum load current detected is 5% of rated current. Current specification is 5(60)A;
The minimum load current is Id, which corresponds to a 3mVrms ADC input signal.
The parameters meet the following relations:

Minimum Detection Load Current Id Rated Current In Maximum Current Imax


ratio to rated current 5% In In 12In
corresponding ADC input signal 3mVrms 60mVrms 720mVrms
actual current 250mA 5A 60A

16 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
4 CALIBRATION

4.1 CALIBRATION METHOD


Normally voltage, current, mean power, phase angle, frequency and so on are regarded as measurement values, while
active energy, reactive energy and so on are regarded as metering value.
Measurement and metering function both need calibration before normal use as shown in below table:

Power Mode Parameter Need Calibration Calibration Method


Normal mode voltage/current offset/gain calibration
power/frequency/phase angle/ power factor X --
THD X --
full-wave energy metering offset/gain/phase angle cali-
bration
fundamental energy metering offset/gain calibration
harmonic energy metering X --
Partial Measurement mode current measurement offset/gain calibration
Detection mode current detection threshold calibration

In typical application of three-phase power meter, voltage, current and full-wave energy must be calibrated. The others can
be calibrated according to actual application, no need to calibrate if no use.
The calibration flow follows the sequence of measurement first then metering. Metering calibration is realized by first
calibrating gain and then calibrating phase angle compensation, only single-point calibration is needed over the entire
dynamic range. Reactive does not need to be calibrated since it is guaranteed by chip design.
Frequency, phase angle and power factor do not need calibration, since their accuracy is guaranteed by chip design.

4.2 CALIBRATION IN NORMAL MODE


The basic functions, such as measurement, metering, harmonic analysis and so on are only active in Normal mode. So
calibration in Normal mode is basic and a must. The related registers need to be configured before calibration. Calibration
flow is as shown in Figure-5.

M90E36A [APPLICATION NOTE] 17


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
Calibration
Initialize

Voltage/Current
Offset Calibration

Voltage/Current
Gain Calibration

Power Offset
Calibration

Energy Gain
Calibration

Phase Angle
Calibration

End

Figure-5 Active Energy Metering Calibration Flow in Normal Mode


4.2.1 Measurement/Metering Startup Command (Configstart/Calstart/HarmStart/AdjStart)
Startup command registers have multiple valid settings for different operation modes.
Startup Register Value Usage Operation
6886H Rower up state It is the value after reset. This state blocks checksum checking error generation
Similar like 6886H, This state blocks checksum checking error generation. Writing with
5678H Calibration
this value trigger a reset to the associated registers.
Checksum checking is enabled and if error detected, IRQ/Warn is asserted and Meter-
8765H Operation
ing stopped.
Other Error Force checksum error generation and system stop.

The default value for these registers is '6886H' after power-on reset. At this time, measurement functions can be started but
metering functions can not. The measurement/ metering functions will be started when related startup registers are set to
'5678H' or '8765H'. If other values are written to these registers, the corresponding measurement/ metering functions will
be disabled, the corresponding checksum and CSxErr bits will be set and the WarnOut pin will output high level.

18 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
Startup Register/ Register Address Range Function Startup on
Register Function CSx / Address
Address (CSx Calculation Range) Reset
ConfigStart / 30H 31H ~ 3AH function configuration CS0 / 3BH --
CalStart / 40H 41H ~ 4CH energy metering calibration CS1 / 4DH not startup
HarmStart / 50H 51H ~ 56H fundamental/ harmonic energy meter- CS2 / 57H not startup
ing calibration
AdjStart / 60H 61H ~ 6EH measurement value calibration CS3 / 6FH startup

When '5678H' is written, the registers resume to their power-on values and metering/measurement functions are started
without checksum check.
When '8765H' is written, the registers do not resume to their power-on values, but checksum will be checked. If the written
checksum is the same as the system self generated checksum, normal metering/measurement functions will be started. If
they are different, metering/measurement functions will not be started, the corresponding CSxErr bits are set and the
WarnOut pin outputs high level. Note that if CS2 is not correct, when the startup register (xxxStart) is 8765H, only harmonic
measurement and metering functions will be disabled. But if CS0, CS1 or CS3 are not correct, all measurement and
metering functions will be disabled.
The written checksum means the value MCU (or other external processor) writes to the addresses 3BH/4DH/57H/6FH
through SPI. The value acquired by MCU reading through SPI is the checksum generated internally. When the startup
register (xxxStart) is 5678H or 8765H, the M90E36A will calculate checksum automatically. As long as there is any register
change, the corresponding CSx value will be updated immediately. So in application the MCU process can be simplified by
reading the CSx registers first to get the correct checksum, then writing the checksum directly back to the CSx registers.

Address Name Bit15 ~ Bit0


01H SysStatus0 Bit15 Bit14 Bit13 Bit12 Bit11 Bit10 Bit9 Bit8
- CS0Err - CS1Err - CS2Err - CS3Err
Bit7 Bit6 Bit5 Bit4 Bit3 Bit2 Bit1 Bit0
URevWn IRevWn - - SagWarn PhaseLos - -
eWn

1 CS0Err: indicates CS0 checksum status


0: CS0 checksum correct (default)
1: CS0 checksum error. The WarnOut pin is asserted at the same time.
2 CS1Err: indicates CS1 checksum status
0: CS1 checksum correct (default)
1: CS1 checksum error. The WarnOut pin is asserted at the same time.
3 CS2Err: indicates CS2 checksum status
0: CS2 checksum correct (default)
1: CS2 checksum error. The WarnOut pin is asserted at the same time.
4 CS3Err: indicates CS3 checksum status
0: CS3 checksum correct (default)
1: CS3 checksum error. The WarnOut pin is asserted at the same time.
In application, it is recommended to set all the startup registers (xxxStart) to 8765H, and timely check the startup registers
(xxxStart) and checksum status indicate bits (CSxErr) in order to judge whether system is in normal operation.

M90E36A [APPLICATION NOTE] 19


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
N Re-initialize
xxxStart == 8765H ?
registers

N
CSxErr == 0 ?

Figure-6 Check the Effectiveness of Register Value


4.2.2 PL Constant Configuration (PL_Constant)
Energy accumulation and metering are usually referenced by energy unit, such as kWh. However, within the M90E36A,
energy calculation or accumulation are based on energy pulse (CF). kWh and CF are connected by Meter Constant (MC,
such as 3200 imp/kWh, which means each kWh corresponds to 3200 energy pulses). The chips PL_Constant is a param-
eter related to MC. One PL_Constant corresponds to 0.01CF. PL_Constant should be configured according to different MC
in application.
The M90E36A provides four energy pulse outputs: active energy pulse CF1, reactive energy pulse or apparent energy
pulse CF2, fundamental energy pulse CF3 and harmonic energy pulse CF4. Their Meter Constants are all set by
PL_Constant in union rather than separately.
The PL_Constant registers consist of the PLconstH[31H] and PLconstL[32H] registers, corresponding to high word and low
word of PL_Constant respectively.
PL_Constant is calculated as below:
PL_Constant = 450,000,000,000 / MC
450,000,000,000: Constant
MC: Meter Constant, unit is imp/KWh, imp/Kvarh or imp/KVA
Example: Calculation of PL_constant
Assume:
Meter Constant MC = 3200
Thus:
PL_constant = 450,000,000,000 / 3200
= 140,625,000 (Hex is 8614C68H)
so the registers are set as below:
PLconstH[31H] = 0861H
PLconstL[32H] = 4C68H

20 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
4.2.3 Metering Method Configuration (MMode0)
The M90E36A can be used in difference systems and metering modes, which can be configured by the MMode0[33H]
register.

Address Name Bit15 ~ Bit0


33H MMode0 Bit15 Bit14 Bit13 Bit12 Bit11 Bit10 Bit9 Bit8
- - I1I3Swap Freq60Hz HPFOff didtEn 001LSB 3P3W
Bit7 Bit6 Bit5 Bit4 Bit3 Bit2 Bit1 Bit0
CF2varh CF2ESV - ABSEnQ ABSEnP EnPA EnPB EnPC

1 I1I3Swap: this bit defines phase mapping for I1 and I3


0: I1 maps to phase A, I3 maps to phase C (default)
1: I1 maps to phase C, I3 maps to phase A
Note: I2 always maps to phase B.
In PCB layout, the M90E36A may be placed on the top layer or bottom layer. The two placements create input
current cross between phase A and C, affecting assembly of the whole meter. The influence can be eliminated by
adjusting this control bit.
Note that the swapping of I1and I3 only changes ADC channels and does not affect the chips internal data
processing.
2 Freq60Hz: grid operating line reference frequency
0: 50Hz (default)
1: 60Hz
The M90E36A is applicable in 50 Hz or 60 Hz power grid. The M90E36A uses different calculation parameters in
data processing according to different grid frequency. To improve the accuracy of measurement and metering,
please set this control bit according to the real power grid frequency.
3 HPFOff: HPF enable control bit
0: enable HPF (default)
1: disable HPF
Besides measuring the voltage/current RMS in 50Hz or 60Hz (AC) power grid, the M90E36A can also measure the
mean current value of DC condition. HPF should be disabled when using DC measurement functions.
4 didtEn: enable integrator for didt current sensor
0: disable integrator for didt current sensor; use CT sampling for current channel (default)
1: enable integrator for didt current sensor; use Rogowski coil sampling for current channel
The M90E36A supports sampling over CT or Rogowski coil. Please set this control bit according to the real current
sampling means. Note that different sampling circuit should be adopted when using Rogowski coil.
5 001LSB: energy register LSB configuration for all energy registers
0: 0.1CF (default)
1: 0.01CF
6 3P3W: connection type for three-phase energy meter
0: 3P4W connection (default)
1: 3P3W connection
The M90E36A uses different phase sequence judgment for different connection. Please set this control bit according
to the real connection type.
7 CF2varh: CF2 pin source configuration
0: apparent energy
1: reactive energy (default)
8 CF2ESV: this bit is to configure the apparent energy computation type when the CF2 pin is set as apparent energy
output. This control bit is also used to configure the apparent energy computation type when calculating power factor
(PF).
0: All-phase apparent energy arithmetic sum (default)

M90E36A [APPLICATION NOTE] 21


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
1: All-phase apparent energy vector sum
9 ABSEnQ: configure the calculation method of total (all-phase-sum) reactive energy and power
0: total reactive energy equals to all-phase reactive energy arithmetic sum (default)
1: total reactive energy equals to all-phase reactive energy absolute sum
10 ABSEnP: configure the calculation method of total (all-phase-sum) active energy and power
0: total active energy equals to all-phase active energy arithmetic sum (default)
1: total active energy equals to all-phase active energy absolute sum
11 EnPA: this bit configures whether Phase A is counted into the all-phase sum energy/power (P/Q/S)
0: Corresponding Phase A not counted into the all-phase sum energy/power (P/Q/S)
1: Corresponding Phase A to be counted into the all-phase sum energy/power (P/Q/S) (default)
12 EnPB: this bit configures whether Phase B is counted into the all-phase sum energy/power (P/Q/S)
0: Corresponding Phase B not counted into the all-phase sum energy/power (P/Q/S)
1: Corresponding Phase B to be counted into the all-phase sum energy/power (P/Q/S) (default)
13 EnPC: this bit configures whether Phase C is counted into the all-phase sum energy/power (P/Q/S)
0: Corresponding Phase C not counted into the all-phase sum energy/power (P/Q/S)
1: Corresponding Phase C to be counted into the all-phase sum energy/power (P/Q/S) (default)

Application note: Common configuration of MMode0


(a) 3P4W, grid frequency 50Hz, MMode0 = 0087H
(b) 3P3W, grid frequency 50Hz, MMode0 = 0185H

4.2.4 PGA Gain Configuration (MMode1)


The MMode1 register is used to configure PGA gain of ADC sampling channel, making chips applicable to meter designs of
different current specifications.

Address Name Bit15 ~ Bit0


34H MMode1 Bit15 Bit14 Bit13 Bit12 Bit11 Bit10 Bit9 Bit8
DPGA_GAIN PGA_GAIN (V3) PGA_GAIN (V2) PGA_GAIN (V1)
Bit7 Bit6 Bit5 Bit4 Bit3 Bit2 Bit1 Bit0
PGA_GAIN (I4) PGA_GAIN (I3) PGA_GAIN (I2) PGA_GAIN (I1)

1 DPGA_GAIN: digital PGA gain for the 4 current channels


00: Gain = 1 (default)
01: Gain = 2
10: Gain = 4
11: Gain = 8
2 PGA_GAIN (V1~V3, I1~I4): analog PGA gain for seven ADC channels
00: 1X (default)
01: 2X
10: 4X
11: N/A

Application note: Configuration principle of PGA gain


(a) Ensure that the ADC channel analog input signal should be within the dynamic range of
0~720mVrms
(b) Configure PGA gain to be the maximum value within the whole dynamic range

22 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
4.2.5 Offset Calibration of Voltage/ Current/ Power
In application, the input signal is often influenced by the interference signal. This interference will enter data processing
module through ADC and high-pass filter, not only producing errors to the voltage/current RMS and power calculation, but
also affecting accuracy of the energy metering. The M90E36A provides offset calibration function to voltage, current and
power, reducing the influence of the interference signal to measurement/metering accuracy.
Every phases voltage/current offset calibration should be proceeded individually. Take phase A for example, the signal
source is: Ub=Uc=Un, Ua=0, Ia=0. The calibration flow of voltage/current offset is as below:
a. Read measurement registers (32 bits). It is suggested to read several times to get the average value;
b. Right shift the 32-bit data by 7 bits (ignore the lowest 7 bits);
c. Invert all bits and add 1 (2s complement);
d. Write the lower 16-bit result to the offset register
Every phases power offset calibration should be proceeded individually. Take phase A for example, the signal source is:
Ua=Ub=Uc=Un, Ia=0. Set the input source to be 0, the calibration flow of power offset is as below:
a. Read measurement registers (32 bits). It is suggested to read several times to get the average value;
b. Calculate: register value x 100,000 / 65,536
c. Right shift the caculated result data by 8 bits (ignore the lowest 8 bits);
d. Invert all bits and add 1 (2s complement);
e. Write the lower 16-bits result to the offset register
The relationship between offset register and measurement register is as below:

Item Data Width Data Align Minimum Unit


voltage rms 32 bits 9 bits 16 bits 7 bits 0.02mV
current rms 32 bits 9 bits 16 bits 7 bits 0.2A
mean power 32 bits 8 bits 16 bits 8 bits 0.00256W
Offset register

The corresponding offset register and measurement value registers are shown as below:

Offset Registers Measurement Value Registers


Address Register Name Address Register Name Address Register Name
Voltage 63H UoffsetA 0D9H UrmsA 0E9H UrmsALSB
67H UoffsetB 0DAH UrmsB 0EAH UrmsBLSB
6BH UoffsetC 0DBH UrmsC 0EBH UrmsCLSB
Current 64H IoffsetA 0DDH IrmsA 0EDH IrmsALSB
68H IoffsetB 0DEH IrmsB 0EEH IrmsBLSB
6CH IoffsetC 0DFH IrmsC 0EFH IrmsCLSB
6EH IoffsetN 0D8H IrmsN1 - -
All-wave 41H PoffsetA 0B1H PmeanA 0C1H PmeanALSB
Power 42H QoffsetA 0B5H QmeanA 0C5H QmeanALSB
43H PoffsetB 0B2H PmeanB 0C2H PmeanBLSB
44H QoffsetB 0B6H QmeanB 0C6H QmeanBLSB
45H PoffsetC 0B3H PmeanC 0C3H PmeanCLSB
46H QoffsetC 0B7H QmeanC 0C7H QmeanCLSB
fundamental 51H PoffsetAF 0D1H PmeanAF 0E1H PmeanAFLSB
power 52H PoffsetBF 0D2H PmeanBF 0E2H PmeanBFLSB
53H PoffsetCF 0D3H PmeanCF 0E3H PmeanCFLSB

M90E36A [APPLICATION NOTE] 23


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
4.2.6 Voltage/ Current Measurement Calibration
Measurement calibration means the calibration of voltage rms (Urms) gain and current rms (Irms) gain. Measurement cali-
bration is the premise of energy metering calibration.
1 Voltage/current offset (Uoffset/Ioffset) calibration:
For calibration method, please refer to 4.2.5 Offset Calibration of Voltage/ Current/ Power. No need of calibration if
the voltage/current offset is very small in general application.
2 Voltage/current gain calibration:
The three phases calibration can be proceeded simultaneously. The signal source is: Ua=Ub=Uc=Un,
Ia=Ib=Ic=In(Ib). The calibration method is as below:
a. Read voltage/current value of the external reference meter, and also read the voltage/current measurement value
from chip registers;
b. Calculate the voltage/current gain:

reference voltage value


Voltage Gain = x 52800
voltage measurement value
reference current value
Current Gain = x 30000
current measurement value
c. Write the result to the corresponding voltage/current gain registers
Note: voltage/current gain calibration is not necessarily proceeded when gain register is the default value. That is,
when the first calibration result is not ideal, there is no need to reset the gain register to the default value. Calibration
can be performed again based on the current value. The formula is as below:
reference voltage value
New Voltage Gain = x existing voltage gain
voltage measurement value

reference current value


New Current Gain = x existing current gain
current measurement value
The corresponding voltage/current gain register and measurement value registers are shown as below:
Gain Register Measurement Value Registers
Address Register Name Address Register Name Address Register Name
Voltage 61H UgainA 0D9H UrmsA 0E9H UrmsALSB
65H UgainB 0DAH UrmsB 0EAH UrmsBLSB
69H UgainC 0DBH UrmsC 0EBH UrmsCLSB
Current 62H IgainA 0DDH IrmsA 0EDH IrmsALSB
66H IgainB 0DEH IrmsB 0EEH IrmsBLSB
6AH IgainC 0DFH IrmsC 0EFH IrmsCLSB
6DH IgainN 0D8H IrmsN1 - -

Application Note:
(a) Voltage rms is unsigned and the minimum unit 1LSB of the UrmsA/UrmsB/UrmsC registers is 0.01V. Only the higher 8
bits of the UrmsALSB/UrmsBLSB/UrmsCLSB registers are valid, the lower 8 bits are always 0, and 1LSB is 0.01/256 V.
(b) Current rms is unsigned and the minimum unit 1LSB of the IrmsA/IrmsB/IrmsC registers is 0.001A; Only the higher 8
bits of the IrmsALSB/IrmsBLSB/IrmsCLSB registers are valid, the lower 8 bits are always 0, and 1LSB is 0.001/256 A.

24 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
Example: Voltage gain calibration
Assume:
The initial value of phase A voltage gain register UgainA is 0CE40H (52800)
Reference meter output voltage is 220.00V
Voltage rms register readout UrmsA = 5BA0H (23456)
The higher 8 bits of voltage LSB register readout UrmsALSB = 4EH (78)
Thus:
voltage measured value = (UrmsA x 0.01) + (UrmsALSB x 0.01 / 256)
= (23456 x 0.01) + (78 x 0.01 / 256)
=234.563 V
voltage gain = 220.00 / 234.563 x 52800 = 49521.88 = 0C172H
So the register can be set to:
UgainA = 0C172H

4.2.7 Energy Metering Calibration


Only active energy is required for energy calibration. There is no need to calibrate reactive energy, the accuracy of which is
guaranteed by chip design. Metering calibration flow is gain first then phase angle. Active energy pulse output (CF1) should
be connected to the pulse input port of the calibration bench during calibration.
Energy metering should be calibrated at In (Ib).
1 Power offset (Poffset/Qoffset) calibration
For calibration method please refer to 4.2.5 Offset Calibration of Voltage/ Current/ Power. No need of calibration if the
power offset is very small in general application.
2 Gain calibration
Every phases gain calibration should be proceeded individually. Take phase A for example, the signal source is:
Ua=Ub=Uc=Un, Ia=In(Ib), Ib=Ic=0, PF=1.0. The calibration method is as below:
a. Read the energy error value e from calibration bench;
b. Calculate the gain;
-
Gain = Complementary ( x 215 )
1+
c. Write the result to the corresponding gain registers.
3 Phase angle calibration. Take phase A for example, the signal source is: Ua=Ub=Uc=Un, Ia=In(Ib), Ib=Ic=0,
PF=0.5L.The calibration method is as below:
a. Read the energy error value ep from calibration bench;
b. Calculate the phase angle error;
AngleError = p Gphase
Gphase is a constant. When grid frequency is 50Hz, Gphase=3763.739. When grid frequency is 60Hz,
Gphase=3136.449
c. Write the result to the corresponding phase angle error registers. The phase angle registers are signed and MSB
of 1 indicates a negative value.
The corresponding gain register and phase angle registers are shown as below:

Address Register Name


Phase A 47H GainA
48H PhiA
Phase B 49H GainB
4AH PhiB
Phase C 4BH GainC
4CH PhiC

M90E36A [APPLICATION NOTE] 25


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
Example: Energy gain and phase angle calculation
The condition is that power factor PF=1.0, current is Ib, energy error is -13.78%, so:
-/(1+)=0.159823707,
gain = int(0.159823707* 2^15)=5237.10=1475H
Write 1475H to the gain register.
After gain calibration, energy error P is 0.95% in the condition of PF=0.5L, current is Ib and
frequency is 50Hz, so:
phase angle = P*3763.739
=0.0095*3763.739=35.75553=24H;
Write 24H to the phase angle register.

4.2.8 Fundamental Energy Metering Calibration


For fundamental energy metering calibration, only gain and offset calibration is needed. There is no need to calibrate phase
angle. Fundamental energy pulse output (CF3) should be connected to the pulse input port of the calibration bench during
calibration.
The startup register for fundamental energy metering calibration is HarmStart [50H]. Calibration related registers are
51H~56H. During calibration, please only start and configure these registers rather than other registers.
Fundamental energy metering calibration is similar to energy metering calibration.
1 Fundamental power offset (PoffsetxF) calibration
For fundamental power offset calibration, only active power error needs to be calibrated per phase individually. Take phase
A for example, the signal source is: Ua=Ub-Uc=Un, Ia=0. For calibration method please refer to 4.2.5 Offset Calibration of
Voltage/ Current/ Power. No need of calibration if the power offset is very small in general application.
2 Fundamental energy gain calibration
Every phases fundamental energy calibration should be proceeded individually. Take phase A for example, the signal
source is: Ua=Ub=Uc=Un, Ia=In(Ib), Ib=Ic=0, PF=1.0. The calibration method is as below:
a. Read the error value e from the external reference meter;
b. Calculate the gain;
-
Gain = Complementary ( x 215 )
1+
c. Write the result to the corresponding gain registers.
The corresponding fundamental energy gain registers are shown as below:

Address Register Name


54H PGainAF
55H PGainBF
56H PGainCF

26 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
Fundamental Energy
Calibration Startup

Fundamental Power
Offset Calibration

Fundamental Energy
Gain Calibration

End

Figure-7 Fundamental Energy Metering Calibration Flow

M90E36A [APPLICATION NOTE] 27


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
4.3 CALIBRATION IN PARTIAL MEASUREMENT MODE
Partial measurement related registers are 14H~1DH. There are no specific startup and checksum registers. Therefore, the
related register should be cleared and configured with initial values before calibration.
4.3.1 Partial Measurement Configuration (PMConfig)

Register
Address Bit15 ~ Bit0
Name
1BH PMConfig Bit15 Bit14 Bit13 Bit12 Bit11 Bit10 Bit9 Bit8
- ReMeasure MeasureStartZX MeasureType - - - -
Bit7 Bit6 Bit5 Bit4 Bit3 Bit2 Bit1 Bit0
- - - - - - - PMBusy

1 ReMeasure: enable another measurement cycle


0: not enable (default)
1: trigger another measurement cycle
The partial measurement module is one-time triggered, that is, once the ReMeasure bit is set, current measurement
is performed one time then turned off. If measurement is required again, the ReMeasure bit should be set again.
2 MeasureStartZX: configure start of measurement
0: Measurement start immediately after the ReMeasure bit is set (default)
1: Measurement start from zero-crossing point after the ReMeasure bit is set
3 MeasureType: indicate the measurement type
0: RMS measurement (default)
1: Mean Value (DC Average) measurement
4 PMBusy: indicate the measure Busy status
0: Measurement done (default)
1: Measurement in progress, Busy
4.3.2 Sampling Cycle Configuration (PMAvgSamples)
The PMAvgSamples[1CH] register is used to configure the partial measurement sampling cycle. The unit is the number of
ADC sampling within a partial measurement cycle. The default grid frequency is 50Hz, so at the ADC sampling rate of 8K,
the default value of PMAvgSamples is 160 (0A0H).
4.3.3 PGAgain Configuration (PMPGA)
The ADC PGAgain Configuration in Partial Measurement mode is similar to the MMode1 configuration in Normal mode.
Address Register Name Bit15 ~ Bit0
17H PMPGA Bit15 Bit14 Bit13 Bit12 Bit11 Bit10 Bit9 Bit8
DPGA_GAIN - - - - - -
Bit7 Bit6 Bit5 Bit4 Bit3 Bit2 Bit1 Bit0
- - PGA_GAIN (I3) PGA_GAIN (I2) PGA_GAIN (I1)

1 DPGA_GAIN: DPGA gain for four current sampling channels


00: Gain = 1 (default)
01: Gain = 2
10: Gain = 4
11: Gain = 8
2 PGA_GAIN(I1~I3): PGA gain for three ADC sampling channels
00: Gain = 1 (default)
01: Gain = 2
10: Gain = 4
11: N/A

28 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
4.3.4 Current Offset Calibration
Considering the influence of the interference signal, Partial Measurement mode also supports current offset calibration.
The calibration method is slightly different from the current offset calibration in Normal mode. Take the rms measurement
for example, the calibration method is as below:
a. Set the input source to be 0;
b. Set MeasureType=1 and ReMeasure=1 to start one current measurement;
c. Read the current registers (16 bits) after measurement completion
d. Repeat step b and c. It is suggested to read many times to get the average value;
e. Invert all bits and add 1 (2s complement);
f. Write the result to the corresponding offset register

Offset Registers Measurement Value Registers


Address Register Name Address Register Name
14H PMoffsetA 18H PMIrmsA
15H PMoffsetB 19H PMIrmsB
16H PMoffsetC 1AH PMIrmsC

Partial Measurement module also provides the measurement value LSB register PMIrmsLSB. This register value has
different definition in rms measurement (AC) and Mean Value measurement (DC) as shown below:

Address Register Name Bit15 ~ Bit0


1DH PMIrmsLSB Bit15 Bit14 Bit13 Bit12 Bit11 Bit10 Bit9 Bit8
- - - - IrmsCLSB
Bit7 Bit6 Bit5 Bit4 Bit3 Bit2 Bit1 Bit0
IrmsBLSB IrmsALSB

In rms measurement, the PMIrmsLSB register value is the LSB of the measurement value. In mean value measure-
ment, this register value is the MSB of the measurement value.
4.3.5 Current Measurement Calibration
The M90E36A only has current measurement function in Partial Measurement mode, so current gain needs to be cali-
brated. Current gain calibration for three phases can be proceeded simultaneously. Take rms measurement for example,
the calibration method is as below:
a. Set MeasureType=0 and ReMeasure=1 to start one current measurement;
b. Read the current value from the signal source (reference meter) and read the value of current registers (16 bits);
c. Calculate the current gain
Current gain = reference current value / current measurement value
d. Partial Measurement mode has no special current gain register, so the calculated result should be saved in MCU
or external memory.
Note that the partial measurement module is enabled in both Normal and Partial Measurement mode. That means the
partial measurement module can measure current value in both modes.

M90E36A [APPLICATION NOTE] 29


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
Partial
Measurement
Statup

Current offset
calibration

Partial
Measurement
Statup

Current gain
calibration

End

Figure-8 Partial Measurement Calibration Flow


4.3.6 Special Application of Partial Measurement Function
In Normal mode, the current rms measurement uses 16 cycles as the measurement period, that means current measure-
ment period is 320ms when grid frequency is 50Hz. If there is a need to measure current at special period, start partial
measurement function in Normal mode, and proceed specific period current measurement by configuring the partial
measurement sampling period register PMAvgSamples. It is noted that employing specific period current measurement
function will not influence the energy metering and parameters measurement functions in Normal mode, since the partial
measurement module is also active in Normal mode.
The partial measurement module can proceed RMS measurement (AC) and Mean Value measurement (DC). The mean
value measurement can also be proceeded in Normal mode by starting partial measurement function.

Application note: How to measure DC current signal


DC measurement function is proceeded by partial measurement module. The flow of DC
measurement is as below:
a. Disable HPF (HPFOff=1)
b. Set MeasureType to 1
c. Set ReMeasure to start current measurement once
d. Detect the IRQ0 pin or the PMBusy bit to judge whether measurement is completed
e. Read the PMIrmsA/PMIrmsB/PMIrmsC/PMIrmsLSB registers to get measurement result
f. Repeat step c~e if to measure again

30 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
4.4 CALIBRATION IN DETECTION MODE
Current detection is realized by low power consumption comparators. The comparator outputs low level when the external
current is lower than the configured threshold; The comparator outputs high level when the external current is higher than
the configured threshold, as shown in Figure-9.

Current Input

Current
DAC IRQ Output
Threshold

Figure-9 Current Detection Principle


4.4.1 Current Detection Module Configuration
Six current threshold comparators can be configured for the current detection module to detect positive and negative
current of three phases. These six threshold comparators can be enabled and disabled by the control bits, as shown below:

Address Register Name Bit15 ~ Bit0


10H DetectCtrl Bit15 Bit14 Bit13 Bit12 Bit11 Bit10 Bit9 Bit8
- - - - - - - -
Bit7 Bit6 Bit5 Bit4 Bit3 Bit2 Bit1 Bit0
- - PDN3 PDN2 PDN1 PDP3 PDP2 PDP1

1 PDN3/2/1: Control bits for negative detector of channel 3/2/1;


0: Detector enable (default)
1: Detector disable
2 PDP3/2/1: Control bits for positive detector of channel 3/2/1;
0: Detector enable (default)
1: Detector disable
Each of the six current threshold comparators has its own register configuration as shown below:

Address Register Name Bit15 ~ Bit0


11H DetectThA Bit15 Bit14 Bit13 Bit12 Bit11 Bit10 Bit9 Bit8
12H DetectThB - CalCodeN
13H DetectThC Bit7 Bit6 Bit5 Bit4 Bit3 Bit2 Bit1 Bit0
- CalCodeP

1 CalCodeN: negative detector threshold, 7-bit width.


7b000-0000 corresponds to minimum threshold Vc=-4.28mV=-3.03mVrms
7b111-1111 corresponds to maximum threshold Vc=12.91mV=9.14mVrms
2 CalcodeP: positive detector threshold, definition is the same as CalCodeN.

M90E36A [APPLICATION NOTE] 31


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
4.4.2 Current Detection Threshold Calibration
Because of the low power consumption consideration and the manufacturing process, the current detection threshold is
different from different chips. Therefore, calibration is needed due to the offset of chips DAC output (less than 5mVrms).
The threshold current range is 2mVrms ~ 4mVrms within which the current detection module (low power consumption
comparator) can detect accurately. It is recommended to proceed system design according to current detection threshold of
3mVrms.
In Detection mode, all registers are not accessible, so the current threshold registers need to be configured in Normal
mode first before entering Detection mode.
Dichotomy is suggested in current detection threshold calibration. The recommended calibration flow is as shown in Figure-
10.

Reference source outputs


current signal that needs
detection (such as 5%Ib)

Th_max = 80H Th_max, Th_min, Th_temp:


Th_min = 00H variable
Th_temp = 40H DetectThx: Threshold register

Switch to Normal mode


DetectThx = Th_temp

Th_min = Th_temp Th_max = Th_temp


Th_temp = (Th_max Th_min) / 2 Th_temp = (Th_max Th_min) / 2

Switch to Detection mode

N N

Y N
Th_max Th_temp == 1 ? IRQ0 output high level? Th_temp Th_min == 1 ?

Y Y

current detection threshold current detection threshold


= Th_temp = Th_min

End

Figure-10 Current Detection Threshold Calibration Flow

32 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
5 FUNCTION REGISTERS CONFIGURATION

5.1 STARTUP CURRENT CONFIGURATION


The registers which related to startup current configuration is shown as below:
Address Register Name Description
35H PStartTh All-phase Active Startup Power Threshold.
36H QStartTh All-phase Reactive Startup Power Threshold.
37H SStartTh All-phase Apparent Startup Power Threshold.
38H PPhaseTh Each-phase Active Startup Power Threshold.
39H QPhaseTh Each-phase Reactive Startup Power Threshold.
3AH SPhaseTh Each-phase Apparent Startup Power Threshold.

1 Due to system interference when current is 0, small signal may be gernerated in current sampling channel, producing
a certain amount of energy and affecting the measurement and metering accuracy. To avoid this, the M90E36A
provides the each-phase startup power configuration/judgment function.
2 PPhaseTh, QPhaseTh and SPhaseTh are used to judge the startup power of each phase (A/B/C). Take active power
for example, when a single phase input power is smaller than the configured PPhaseTh value, the input active power
of that phase will be set to 0 by force, that means input to the next process is 0. Otherwise the signal will be streamed
to the next process.
Note that the threshold are configured separately to active(P), reactive(Q) and apparent (S). The compared value is
(|P|+|Q|).
3 PStartTh, QStartTh and SStartTh are used to judge all-phase startup power. Take active power for example, when all-
phase-sum power is less than the configured PStartTh value, energy accumulation will not start. Otherwise energy
accumulation will start.
4 Calculation methods of the two register groups are the same. The formula is as below:
Register value = N / 0.00032, (N is the configured power threshold).

Example: Startup Current Configuration


Assume:
meter voltage is 220V, current specification 5(100)A, active startup current is 0.1%
Considering the accuracy of current measurement in small-current state, it is recommended to configure the all-
phase startup current threshold to be 50% of startup current (also can configure based the actual conditions).
Assume the startup threshold of each-phase power is configured to be 10% of startup current.
so:
All-phase Active Startup Power Threshold = 3 x 5 x 0.1% x 50% x 220 = 1.65W
Each-phase Active Startup Power Threshold = 5 x 0.1% x 10% x 220 = 0.11W
Register values are:
PStartTh[35H] = 1.65 / 0.00032 = 5156 = 1424H
PPhaseTh[38H] = 0.11 / 0.00032 = 344 = 158H

M90E36A [APPLICATION NOTE] 33


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
0 0 Total Active
Power Threshold Total Active Power
1 Energy Metering
|P|+|Q|>
PPhaseTh? ABS >
3 phases + PStartTh?

Phase Active 1
A/B/C
Power from DSP
1
0 Phase Active
0
Energy Metering
0 0

0 0 Total Reactive
Power Threshold Total Reactive Power
1 Energy Metering
|P|+|Q|>
QPhaseTh? ABS >
3 phases + QStartTh?

Phase Reactive 1
A/B/C
Power from DSP
1
0 Phase Reactive
0
Energy Metering
0 0

Power Threshold 0 0 Total (arithmetic


Total Apparent Power
|P|+|Q|> 1 sum) Apparent
SPhaseTh? Energy Metering
ABS >
3 phases + SStartTh?

Phase Apparent 1
A/B/C
Power from DSP
1
0 Phase Apparent
0
Energy Metering
0 0

Figure-11 Metering Startup Handling

5.2 SAG FUNCTION


Sag detect function is provided in M90E36A. The threshold of sag detection is configured through the SagTh register
(08H). All three voltage phases use the same threshold. The threshold equation is as below:
Vth 100 2
SagTh =
2 Ugain / 32768

Vth: the voltage threshold to be configured;


Ugain: the gain after calibration
The default value of the SagTh register (08H) is 0000H. It is suggested to set the SagTh register (08H) appropriately even
if the Sag function is not used in application. The reasons are as follows:
In normal mode, all measurement values are calculated based on the average cycle out of 16 voltage cycles. And the
voltage cycle makes use of the internal zero-crossing signal, which is different from the output on the ZX2 / ZX1 / ZX0 pins.
In a 3P4W system, this internal zero-crossing signal is based on phase A voltage Ua firstly. If phase A is in sag, phase C
voltage Uc is switched. If phase A and phase C are both in sag, phase B voltage Ub is switched. In a 3P3W system, the
internal zero-crossing signal is based on phase A voltage Uab firstly. If phase A is in sag, phase C voltage Ucb is switched.
In either 3P4W and 3P3W application, if all phases are in sag, the average cycle of 16 voltage cycles is calculated in accor-
dance with the configured reference frequency (50 Hz or 60 Hz), i.e. 320ms for 50Hz system and 266.7ms for 60Hz
system.
However, for energy meter or power instrument with auxiliary power supply, if the SagTh register is not configured (the
default power-on value is 0), the M90E36A will not enter sag even if there is no signal on voltage circuits due to the inter-
ference noise. Thus the wrong average cycle might be used to calculate measurement values such as current rms.

34 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
5.3 RESERVED REGISTER/ ADDRESS AND RESERVED BITS
5.3.1 Reserved Register/ Address
The M90E36A has many reserved registers and non-listed address areas besides the registers listed in datasheet. These
reserved registers are not open to users. Please dont operate on the address outside of the datasheet. Access to reserved
register/address needs special operation because normal operation can not make change or impact to these register
values.
5.3.2 Reserved Register Bits
Some fields of defined registers in the datasheet are labeled as reserved. In application, all the reserved bit fields shall be
written with 0 when those bits have to be written, and shall be masked out (hence ignored) upon read.
5.3.3 Reserved Bits in the FUNC_EN1 Register
The most registers reserved bits will return fixed 0 upon read. There is only one exception: the FUNC_EN1 register (04H).
This register has three reserved bits. Those bits may return non-fixed value upon read. Users need to mask out or ignore
those bits when using the read-back value of this register. Please note that the non-fixed return value does not indicate any
abnormal working condition of the internal hardware logic. Users just need to ignore those bit fields.

Address Name Bit15 ~ Bit0


04H FuncEn1 Bit15 Bit14 Bit13 Bit12 Bit11 Bit10 Bit9 Bit8
INOv1En INOv0En Reserved Reserved THDUOvEn THDIOvEn DFTDone Reserved
Bit7 Bit6 Bit5 Bit4 Bit3 Bit2 Bit1 Bit0
RevQchg- RevQch- RevQchg- RevQchg- RevPchg- RevPch- RevPchg- RevPchg-
TEn gAEn BEn CEn TEn gAEn BEn CEn

M90E36A [APPLICATION NOTE] 35


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
6 TEMPERATURE COMPENSATION
The M90E36A itself embodies good temperature characteristic. Considering that the external components might be
affected by temperature in application, the M90E36A also provides compensation function for external temperature drift.
A series of special registers should be configured for temperature compensation. These registers are located in special
addresses, and access to these registers should be strictly carried out as the following.

6.1 ON-CHIP TEMPERATURE SENSOR CONFIGURATION


The M90E36A provides a built-in temperature sensor. Due to the manufacturing process, the temperature sensor might be
somewhat different for different chips. Therefore the on-chip temperature sensor should be configured before temperature
compensation.
The configuration method is as below:
a. Write AA55H to address 2FDH
b. Write 5122H to address 216H
c. Write 012BH to address 219H
d. Write 0000H to address 2FDH
Read the Temp[0FCH] register directly to get the current temperature after configuration completed. Please note that, the
temperature sensor will sense the temperature of the chip and it may have a few degrees of difference between the chip
junction temperature and ambient temperature.

6.2 TEMPERATURE COMPENSATION BASED ON ADC SAMPLING CHANNEL


The temperature compensation method is as below:
a. Write AA55H to address 2FCH
b. Write the temperature coefficient to be compensated to address 270H
c. Write the fiducial point temperature of the temperature coefficient to address 27BH
d. Write 0000H to address 2FCH
The reference point temperature of the temperature coefficient is generally configured to be the temperature in calibration.
That is, in calibration, read the Temp[0FCH] register first to get the current temperature and then save it as the reference
temperature.

Address Register Name Bit Read/Write Default Value Description


270H TempCompGain 15:8 Read/Write 0 Reserved bit. Readout value is 0.
7:0 - - Temperature compensation coefficient, bit 7 is the sign
bit, unit is ppm/
27BH TempCompRef 15 Read/Write 0 1: enable temperature compensation
0: disable temperature compensation
14:9 - - Reserved bit. Readout value is 0
8:0 Read/Write 19H The reference temperature of the temperature coeffi-
cient, bit 8 is the sign bit, unit is

36 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
Example:
Test data before temperature compensation is as below:

0.08
Error

0.06

0.04

0.02

0 Temp
-50 0 50 100
-0.02

-0.04

-0.06

-0.08

Error Linear Compesation

After linearization:
The reference temperature (temperature in calibrating) is 25 , metering error is 0.0000%;
The error at 85 point is 0.06%
So the temperature coefficient is calculated as below:
(0.06% - 0.0000%) / (85 - 25 ) = 10ppm/
The temperature coefficient to be compensated is -10ppm/ . Register configuration is as below:
a. Write AA55H to address 2FCH
b. Write FFF6H to address 270H
c. Write 8019H to address 27BH
d. Write 0000H to address 2FCH

M90E36A [APPLICATION NOTE] 37


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
6.3 TEMPERATURE COMPENSATION BASED ON REFERENCE VOLTAGE
On-chip high-precision reference voltage is provided with excellent low temperature coefficient. But in application, what
should be considered is the temperature drift of the whole system. Therefore, the M90E36A specially provides temperature
compensation based on reference voltage to minimize temperature drift caused by the on-board components.
Note that, as voltage and current ADC sampling adopt the same reference voltage, compensation on the reference voltage
will bring double effect on power and energy.
Temperature compensation on reference voltage is proceeded with every 8 as a segment. In application, it is suggested
to test on a small batch of components from the same lot to get the best temperature compensation coefficient. And then
use this compensation coefficient as a fixed value to be written directly to register in production.
The temperature compensation method is as below:
a. Write AA55H to address 2FDH
b. Write the reference voltage coefficient of segmented compensation to addresses 202H~209H
c. Write the curvature of segmented compensation curve to address 201H
d. Write 0000H to address 2FDH
In normal condition, reference voltage is 1200mV. The unit of reference voltage compensation is 0.020mV in this compen-
sation method.
The default value of the corresponding compensation registers is the ideal value in chip design. In application, only incre-
mental adjustment is needed.

Address Register Name Bit Read/Write Default Value Description


201H BGCurveK 15:4 - - Reserved bit, readout value is 0
3:0 Read/Write 0 Reference voltage temperature compensation curve.
The bit3 is assumed as sign bit, range is -8 to +7.
Scaling factor = 1+ register value*1/8.
So the scaling factor will be from 0, with step of 1/8, all the
way to 1+7/8.
202H BG_TEMP_P12 15:8 Read/Write 0 Compensation coefficient on the 25 temperature point
7:0 Read/Write 1 Compensation coefficient on the 17 temperature point
203H BG_TEMP_P34 15:8 Read/Write 6 Compensation coefficient on the 41 temperature point
7:0 Read/Write 2 Compensation coefficient on the 33 temperature point
204H BG_TEMP_P56 15:8 Read/Write 25 Compensation coefficient on the 57 temperature point
7:0 Read/Write 13 Compensation coefficient on the 49 temperature point
205H BG_TEMP_P78 15:8 Read/Write 53 Compensation coefficient on the 73 temperature point
7:0 Read/Write 39 Compensation coefficient on the 65 temperature point
206H BG_TEMP_N12 15:8 Read/Write 16 Compensation coefficient on the 1 temperature point
7:0 Read/Write 6 Compensation coefficient on the 9 temperature point
207H BG_TEMP_N34 15:8 Read/Write 54 Compensation coefficient on the -15 temperature point
7:0 Read/Write 34 Compensation coefficient on the -7 temperature point
208H BG_TEMP_N56 15:8 Read/Write 117 Compensation coefficient on the -31 temperature point
7:0 Read/Write 79 Compensation coefficient on the -23 temperature point
209H BG_TEMP_N78 15:8 Read/Write 205 Compensation coefficient on the -47 temperature point
7:0 Read/Write 159 Compensation coefficient on the -39 temperature point

38 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
7 HARMONIC ANALYSIS

7.1 DFT ENGINE


The built-in DFT computation engine supports 2nd-32nd order harmonic analysis function for 6 channels. The calculation
error will be enlarged when input signal is small. To address this issue, a prescaler is designed and placed before the DFT
engine to amplify the signal to be calculated.
The designed ADC sampling rate is 8kHz. Harmonic analysis adopts 4096 sampling points for DFT computation which
takes around 0.5s once.
Considering there are many DFT computation outputs, the DFT computation engine is closed by default to ensure external
MCU can read DFT data from the same calculation. The engine needs to be enabled to startup, and it will automatically
shut off after completing calculation for one time. Calculation results are stored in registers. External MCU can read these
registers to get calculation results through SPI interface.
Harmonic measurement accuracy is guaranteed by chip design after voltage/current calibration.
The control registers of DFT computation engine is as below:

Address Register Name Bit Read/Write Default Value Description


1D0H DFT_SCALE 15 Read/Write 0 0: Enable Hanning window
1: Disable Hanning window
14:13 Read/Write 0 Voltage scale for phase C.
12:11 Read/Write 0 Voltage scale for phase B.
10:9 Read/Write 0 Voltage scale for phase A.
8:6 Read/Write 0 Current scale for phase C.
5:3 Read/Write 0 Current scale for phase B.
2:0 Read/Write 0 Current scale for phase A.
1D1H DFT_CTRL 15:1 - 0 Reserved bit.
0 Read/Write 0 0: Disable DFT engine
1: Enable DFT engine

1 The function of Hanning window is to bring periodicity to ADC sampling signal in DFT computation to achieve the exact
calculation result. Please enable Hanning window in general application.
2 Voltage scale, InputGain= 2^Scale
00: Gain = 1
01: Gain = 2
10: Gain = 4
11: Gain = 8
3 Current scale, InputGain= 2^Scale
000: Gain = 1
001: Gain = 2
010: Gain = 4
011: Gain = 8
100: Gain = 16
101: Gain = 32
110: Gain = 64
111: Gain = 128
4 DFT engine switch: DFT computation engine is enabled after setting the DFT_CTRL bit. This bit will be cleared auto-
matically after completing calibration for one time. In application, this bit can be used to judge whether DFT compu-
tation is completed.

M90E36A [APPLICATION NOTE] 39


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
DFT application is as below:
a. Set DFT computation engine and write 2A49H to the DFT_SCALE [1D0H] register (Assume gain of voltage and current is two)
b. Start DFT computation engine and write 001H to the DFT_CTRL [1D1H] register
c. Check DFT_CTRL. If DFT_CTRL=0, DFT computation is completed (about 0.5s)
d. Read register value and get harmonic component and fundamental voltage/current value after transition

Harmonic Component (%) = Register Value / 163.84


Register address 100H~1BFH

Register Value x 3.2656


Fundamental Current =
2Scale x 1000
Register address 1C0H, 1C2H, 1C4H

Register Value x 3.2656


Fundamental Voltage =
2Scale x 100
Register address 1C1H, 1C3H, 1C5H
For description of the related registers, please refer to datasheet.

Example:
Assume:
Meter s nominal voltage is 220V, nominal current is 5A
The signal source outputs 10% of 5th order harmonic component for phase A voltage and
40% of 5th order harmonic for phase A current.
Register values are as follows after DFT computation engine:
[103H] = 0671H (1649)
[123H] = 19C9H (6601)
so the measured harmonic component is :
Voltage 5th order harmonic component = 1649 / 163.84 = 10.0647 (means 10.0647%)
Current 5th order harmonic component = 6601 / 163.84 = 40.2893 (means 40.2893%)

In application, the THD+N threshold for three-phase voltage and three-phase current can be configured. The M90E36A can
judge whether the THD+N value is greater than the configured threshold by checking the corresponding status bits or the
IRQ output signal. If the THD+N value is greater than the configured threshold, DFT computation engine can be started,
and analysis and recording can be processed to the harmonic signal.

40 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
Set THD+N monitor
threshold

N
Check THD+N detector

Start DFT function

N
Detect DFT completion
identification

Read the DFT computation


result directly

Figure-12 Harmonic Analysis Application Flow

M90E36A [APPLICATION NOTE] 41


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
7.2 OBTAIN HARMONIC ANALYSIS OF ABOVE 32ND
Harmonic analysis of 32nd to 42nd can be obtained as below:
1. Read the [1D7H] register (divided by 1024, current frequency can be achieved);
2. Write the value of the [1D7H] register to the 1D4H register;
3. Calculate the value of the [1D7H] register *15.872, then convert to hex. Write the high word to the [1D2H] register
and the low word to the [1D3H] register;
4. Write 0x101 to the [1D1H] register to start DFT calculation;
5. Read the 33rd - 42nd order component of each phases voltage and current in the 2nd - 11th registers when DFT
calculation is complete after 0.5s (typical).
For example, if the current frequency is 50Hz, the value of the [1D7H] register is 51200 (or 0xC800). Write 0xC800 to the
[1D4H] register. Meanwhile 51200*15.872=812646.4=0xC 6666, so write 0x0C to the [1D2H] register and 0x6666 to the
[1D3H] register.
Of course, if not consider the influence of frequency changes to harmonic analysis, step 3 can be simplified as: write 0xC to
the [1D2H] register and 0x6666 to the [1D3H] register.
It should be pointed out that because of the multiplex of some registers, the 90E36/36A can not provide harmonic analysis
of the 2nd - 32nd and the 33rd - 42nd simultaneously. The common practice is to read the 2nd - 32nd harmonic analysis
first, and then read the 33rd - 42nd harmonic analysis. Hence their corresponding intervals are different.
The THD data of each phases voltage/current is analyzed based on current calculated harmonic.
2nd - 32nd harmonic analysis:

31

| X (k ) |
k =1
2

THDI = 100%
| X ( 0) | 2
33rd - 42nd harmonic analysis:

41

| X (k ) |
k = 32
2

THD II = 100 %
| X (0) |2
So the THD for 2nd - 42nd order harmonic:

THDT = THDI 2 + THDII 2

42 M90E36A [Application Note]


Atmel-46104A-SE-M90E36A-ApplicationNote_050514
Revision History

Doc. Rev. Date Comments

46104A 5/5/2014 Initial release.

X X X X
Atmel Corporation 1600 Technology Drive, San Jose, CA 95110 USA T: (+1)(408) 441.0311 F: (+1)(408) 436.4200 | www.atmel.com

2014 Atmel Corporation. All rights reserved. / Rev.: Atmel-46104A-SE-M90E36A-ApplicationNote_050514.

Atmel, Atmel logo and combinations thereof, Enabling Unlimited Possibilities, and others are registered trademarks or trademarks of Atmel Corporation or its
subsidiaries. Other terms and product names may be trademarks of others.

DISCLAIMER: The information in this document is provided in connection with Atmel products. No license, express or implied, by estoppel or otherwise, to any intellectual property right
is granted by this document or in connection with the sale of Atmel products. EXCEPT AS SET FORTH IN THE ATMEL TERMS AND CONDITIONS OF SALES LOCATED ON THE
ATMEL WEBSITE, ATMEL ASSUMES NO LIABILITY WHATSOEVER AND DISCLAIMS ANY EXPRESS, IMPLIED OR STATUTORY WARRANTY RELATING TO ITS PRODUCTS
INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTY OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, OR NON-INFRINGEMENT. IN NO EVENT
SHALL ATMEL BE LIABLE FOR ANY DIRECT, INDIRECT, CONSEQUENTIAL, PUNITIVE, SPECIAL OR INCIDENTAL DAMAGES (INCLUDING, WITHOUT LIMITATION, DAMAGES
FOR LOSS AND PROFITS, BUSINESS INTERRUPTION, OR LOSS OF INFORMATION) ARISING OUT OF THE USE OR INABILITY TO USE THIS DOCUMENT, EVEN IF ATMEL HAS
BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. Atmel makes no representations or warranties with respect to the accuracy or completeness of the contents of this
document and reserves the right to make changes to specifications and products descriptions at any time without notice. Atmel does not make any commitment to update the information
contained herein. Unless specifically provided otherwise, Atmel products are not suitable for, and shall not be used in, automotive applications. Atmel products are not intended,
authorized, or warranted for use as components in applications intended to support or sustain life.

SAFETY-CRITICAL, MILITARY, AND AUTOMOTIVE APPLICATIONS DISCLAIMER: Atmel products are not designed for and will not be used in connection with any applications where
the failure of such products would reasonably be expected to result in significant personal injury or death (Safety-Critical Applications) without an Atmel officer's specific written
consent. Safety-Critical Applications include, without limitation, life support devices and systems, equipment or systems for the operation of nuclear facilities and weapons systems.
Atmel products are not designed nor intended for use in military or aerospace applications or environments unless specifically designated by Atmel as military-grade. Atmel products are
not designed nor intended for use in automotive applications unless specifically designated by Atmel as automotive-grade.